<Mis012[m]>
well... personally I wouldn't mind modelling the hierarchy
<Mis012[m]>
but it's often not done
<DavidHeidelberg[m]>
ok, latest `-next` gives me `platform 5100000.mdp: Fixing up cyclic dependency with 4700000.dsi` and few like `amba 1a01000.etb: Fixing up cyclic dependency with replicator`, in the end. When reverting "removing of amba bus" it gets OK
<DavidHeidelberg[m]>
so, it seems like it "requires" amba node to be there?
<DavidHeidelberg[m]>
or I could make `amba` `apb@_first_address_of_mmc_present_on_APB`, could that make sense?
<Mis012[m]>
wait...
<Mis012[m]>
if it has compatible of `primecell`, then it definitely should be under an amba node xD
<DavidHeidelberg[m]>
trick is, that coresight is directly in soc node
<DavidHeidelberg[m]>
generally then coresight should be also under apb@ node
<Mis012[m]>
I think it is on some SoCs at least
<DavidHeidelberg[m]>
ok, keeping it in `apb@` node instead of `amba` also works
<z3ntu>
Hi, I was wondering what the current best practices are for pinctrl node & label naming. According to bindings the node name should end with -pins but I often also see -active (or just _a) -default and whatever other suffixes
<z3ntu>
and to make it even more complicated newer bindings seem to use -state as suffix
<Mis012[m]>
-active makes sense when paired with -sleep
<Mis012[m]>
which is something mainline drivers usually don't bother with
<Mis012[m]>
-default is like "downstream did -active and -sleep here, but meh"
<Mis012[m]>
haven't really seen the other variants
<DavidHeidelberg[m]>
z3ntu: best would be to ask Rob Herring on #devicetree
<DavidHeidelberg[m]>
lumag: after solving my mess, your patches seems to be working as they should (also the Nexus 7 is sometimes slightly unstable, so it maybe because of that it didn't worked first time)